LANDSCAPE ECOLOGY A Top-Down Approach - Chapter 6 pdf

Landscape and Edge Effects on Population Dynamics: Approaches and Examples Lennart Hansson CONTENTS Introduction Metapopulations Types of Dynamics Spatial Distribution Effects of Habitat

Landscape and Edge Effects on Population Dynamics: Approaches and Examples

Lennart Hansson

CONTENTS

Introduction

Metapopulations

Types of Dynamics

Spatial Distribution

Effects of Habitat Juxtaposition

Population Effects

Community Interactions

Behavior at Edges

Effects at Various Spatial Scales

Landscape Complementation or Supplementation

at Short Distance

Effects at the True Landscape Scale

Effects at Larger Spatial Scales

Some Limited Generalizations

Evolutionary and Historical Background

Heterogeneity or Landscape Ecology?

How to Test Landscape Effects

Landscape Ecology and Conservation

Acknowledgments

Introduction

Until the early 1980s, population dynamics were almost always modeled, conceptually or mathematically, for a homogeneous area without any edge However, most habitat patches are small, particularly with regard to

wide-ranging vertebrates, and edge effects are indeed common It is surprising that field ecologists did not object to this unrealistic representation of nature until

10 to 15 years ago There was a gradual change in the late 1980s and a more pronounced one in the 1990s, particularly with the re-apprehension of the metapopulation approach And during the 1990s the theoreticians have again found a new playground, now starting to model effects of environmental het-erogeneity However, the early observations about “suprahabitat” or land-scape effects (Levins 1969; Anderson 1970; Hansson 1977; Wegner and Merriam 1979) were usually not sympathetic to traditional theory In this chapter I will look at dynamics in heterogeneous areas (usually the landscape scale, as discussed later, with two or several ecosystems) from a field biolo-gist’s rather than a theorist’s point of view

Landscape composition has various influences on populations of mobile organisms So far, mainly animals have been considered, but effects on plant propagules ought also to be examined Furthermore, nonrandomly distrib-uted mobile herbivores may have pronounced local effects on sedentary plant populations The use of landscapes by fungi or microorganisms has hardly been considered at all Thus, this discussion will mainly be limited to animals

Environmental heterogeneity can have a multitude of effects on popula-tions A list of sensitive population attributes, not always mutually indepen-dent, may include:

a Distributions Population may shift from uniform to clumped distributions depending on habitat patch size or juxtaposition

b Persistence The longevity of local populations is dependent on the size of the inhabited area, dispersal routes and distances to other populations

c Type of dynamics Important predators or pathogens, or the physical environment, may affect dynamics differently in vari-ous spatial contexts

d Regulation or density level Populations may lose individuals at edges, affecting mean density level or setting an equilibrium

e Habitat overflow Certain habitats allow higher recruitment than others, causing population expansion or source–sink dynamics

in a landscape

f Habitat interdependence Individuals from one habitat may glean necessary or additional resources in a neighboring habitat whereby population growth rates are impacted Habitat juxta-position may be necessary or beneficial

g Connectivity between habitat patches will affect population sizes and existence

These landscape effects are interwoven for most populations Here, I will present types of populations that are affected by environmental heterogene-ity, and provide some examples, mainly from research at Uppsala, Sweden,

or in Scandinavia generally The ecological setting is boreal to temperate environments I will finally try to make generalizations regarding common effects of environmental heterogeneity, evolution of landscape use, and the future development of landscape ecology

Metapopulations

The idea of random extinction and colonization of subpopulations in isolated habitat patches in the theory of metapopulation dynamics by Levins (1969, 1970) was a pronounced break with contemporary theory, but there was a lapse of some 20 years before it was fairly generally accepted It was, thus, the first approach to population dynamics in a heterogeneous environment (Fig -ure 6.1) Now it is at the center of much conservation theory (Hanski 1994; Gilpin and Hanski 1995) However, in order to be general it contains clearly unrealistic components as similar-sized patches at similar distances Its pre-dicted equilibrium populations have actually been observed in fairly few instances (Harrison 1991) Indeed, many cases of metapopulations, the con-cept taken very widely, have turned out to be satellite or sink populations to larger source areas (Pulliam 1988) or declining populations, particularly of endangered species (Hanski 1996)

I will demonstrate some problems of applying the metapopulation theory

to any subdivided population, using as an example the pool frog studied by Sjögren Gulve (1991, 1994) and Sjögren Gulve and Ray (1996) close to Upp-sala in south-central Sweden At its northern distribution border, it occurs in many separate subpopulations, breeding in pools isolated from the Baltic by land uplift The total number of permanently or temporarily occupied pool populations was defined as a metapopulation by Sjögren Gulve However, the pools were not identical, as in Levin´s model, and differed in several rec-ognizable features: occupied pools had higher water temperature in spring and were close to other occupied pools There were both deterministic and more “stochastic” extinctions, the former being due to pool succession or drainage However, also the latter type of extinctions were predictable in the sense that they occurred close to the primary deterministic extinctions Thus, dispersal was a critical element in the metapopulation system, declining with distance from source populations Dispersing frogs were moving through forests between the pools and also overwintered in moist forest sites This dis-persal and wintering was evidently negatively affected by large-scale drain-age and clear-cutting, creating an environment too dry for successful dispersal (Sjögren Gulve and Ray 1996) Thus, although this particular metap-opulation appeared to be at equilibrium (Sjögren 1991), the colonizations and

extinctions were not random processes, but were strongly affected by the matrix between habitat patches

I suspect that similar conditions occur within most “metapopulations” (see, e.g., Kindvall 1996 below) and that effects from the matrix have to be considered for most subdivided populations The metapopulation concept is strongly connected to purely stochastic processes and if such circumstances

do not apply, the term “subdivided populations” and identifications of matrix effects might be preferable

Types of Dynamics

Variations in dynamics include equilibrium populations, outbreaks, cyclic, and chaotic populations (e.g., Hassel and May 1990) Populations may also move towards extinction Populations in different parts of a geographical range can show different types of dynamics and local populations can change from one type of dynamics to another Most examples come from folivorous insects (Berryman 1981), but similar variation has been observed among foli-vorous mammals (Pimm and Redfern 1988) Both regional and temporal vari-ation may be due to landscape composition

I will first discuss landscape effects on the regionally different fluctuation patterns in Scandinavian small rodents A geographical gradient in fluctua-tion patterns in Scandinavia is well established (Hansson 1971; Hansson and Henttonen 1985; Hanski et al 1993; Turchin 1993; Björnstad et al 1995), with more heavily fluctuating populations in the north At population peaks in northern Scandinavia a wider spectrum of habitats is also utilized than in south Scandinavia (and in central Europe) The level of fluctuation is posi-tively related to the amount of snow cover (Hansson and Henttonen 1985) The explanation suggested is that the snow prevents predation by large gen-eralistic predators by isolating the small rodents below the snow and by

FIGURE 6.1

Particularly emphasized subjects in

the short history of a landscape

ecol-ogy centered on populations There

is a conceptual line running from

metapopulations to matrix effects

over fragmentation and connectivity

while dynamics and edge effects

have been made more separate

sub-jects It should be possible to come to

a joint understanding of matrix and

edge effects.

diminishing numbers of alternative prey In areas poor in snow, generalists may stabilize rodent numbers by switching between prey species In snow-rich areas, predators living under the snow get an advantage, but they consist only of specialized mustelids that will overexploit the rodent populations and cause the pronounced population cycles in voles that are so typical of northern snowy regions (Korpimäki et al 1991) The snow is thus the factor that is causing homogenization of various landscapes for different lengths of time These observations and explanations have changed the way people have been looking at rodent dynamics (Stenseth and Ims 1993) and, although not globally accepted, this hypothesis is supported by various simulation models (Hanski et al 1991; Björnstad et al 1995; Hanski and Korpimäki 1995; Hanski and Turchin, unpublished)

Temporal variations in population fluctuations have also occurred in Scan-dinavia (Hansson 1992, 1994; Hörnfeldt 1994; Lindström and Hörnfeldt 1994; Hanski and Henttonen 1996) Small rodents were strongly cyclic in central Scandinavia in the 1960s and 1970s, but gradually lost their high degree of cyclicity in the 1980s, and these latter populations appeared almost stable with only seasonal density variations in the early 1990s Conversely to the geographical gradient, there is not any generally accepted explanation of the temporal change, but the following one is in agreement with recognizable changes in the landscape

Extensive clear-cutting started in the 1950s in central Scandinavian forests, perhaps particularly in central-northern Sweden The most productive for-ests were cut first, and the clear-cuts were then really huge At the same time,

a lot of agricultural land was abandoned in the same region Early

succes-sional, grassy patches on forest ground are prime habitats for Microtus voles,

as are fairly early phases of overgrowing farmland (Stenseth and Lidicker 1992) From the 1970s onwards smaller clear-cuts were taken up on less

pro-ductive land, old clear-cuts were reforested and no longer of any use to

Micro-tus (Hansson 1994), while abandoned fields went into stages dominated by

coarse grasses that are less profitable to Microtus Thus, there were large and

close patches of sheltering and nutrient-rich vegetation in the 1960s and dis-persal between patches was simple High reproduction and population growth rates then prolonged the time before weasels could overtake the

Microtus populations, but when they finally did, they depressed the voles to

very low-density levels In the 1990s, productive patches were small and

dis-tant, and small or transient, slow-growing Microtus populations were less

able to colonize new patches Weasel populations did not reach as high den-sities as previously (Hansson, unpublished) and generalistic predators may have had great impact during the summer dispersal Many vole populations

now also declined during summer Clethrionomys voles in adjoining forests,

which have lower population growth rates, may have fluctuated due to switching between prey species by weasels and other rodent predators

(Heikkilä et al 1994) However, Clethrionomys voles also prefer the most

lux-uriant forests that were cut already early on

Similar observations and suggestions were made for disappearance of vole cycles at the well-studied Wytham Wood near Oxford by Richards (1985) For

an alternative, or rather complementary, explanation involving chaotic

dynamics, but also interactions between clear-cut–living Microtus and forest-living Clethrionomys via predators, see Hanski and Henttonen (1996).

Similar large geographical and temporal variation has been observed in the population dynamics of insects, e.g., the winter moth Epirrita autumnata, showing pronounced population outbreaks in certain areas, but not in others,

in northern Fennoscandia and never in the European Alps (Tenow and Nils-sen 1990; Berryman 1996) Evidently, accumulation of cold air in valley bot-toms kills the insects’ eggs while the insects may thrive on valley slopes and there exhibit pronounced population cycles

Fragmentation of, e.g., forests may or may not cause declining populations and extinctions, depending on movement characteristics of focal organisms There may be only a statistical sampling effect or true isolation effects (Andrén 1994, 1996) Dispersal ability is thus very important for the final out-come The European red squirrel displays isolation effects for forest patches surrounded by agricultural fields, but not when surrounded by clear-cuts (Andrén and Delin 1994, Delin and Andrén 1997); varying behavioral reac-tions to the matrix type are evidently crucial

Observations of population extinctions in heavily fragmented habitats are commonplace In Scandinavia, the boreal forest was fragmented by forest fires in the pristine state, but moist or wet forests did not burn or burned very seldom (Hansson 1997) Many species inhabiting such moist forests show lit-tle dispersal Species adapted to these fire refugia are therefore very sensitive

to modern forest management with extensive clear-cutting in smaller or larger blocks Examples of such organisms with populations at present mov-ing towards extinction are several lichen species dependmov-ing on a humid

envi-ronment, e.g., the large lichen Usnea longissima (Esseen et al 1997).

Spatial Distribution

The distribution of individuals in a population is commonly expressed as even, random, or clumped, usually after statistical tests in relation to a Pois-son distribution These distributions may be affected by landscape composi-tion, at least when populations fluctuate strongly

Small rodents on clear-cuts in fairly stable south Scandinavian populations show a pronounced clumpiness while cyclic small rodent populations in north Scandinavia demonstrate generally even or random distributions (Hansson 1990) However, in the southern region there were areas with devi-ating distributions Snow depth explained a large proportion of the variation between the southern and northern populations (Hansson 1989) The low level of clumping in the north is supposed to be related to an easy dispersal

under a landscape-wide snow cover (Hansson 1992) Even in noncyclic pop-ulations exposed to heavy predation, dispersal may be more or less easy, depending on the density or structure of the vegetation

These animals still change habitats or microhabitats during a population

cycle On central Swedish clear-cuts, the vole M agrestis moved from wet

parts of clearcuts, with dense grass cover at the peak and early decline to

boulder fields with less food at the same time as weasels (Mustela nivalis)

invaded the grassy parts (Hansson, unpublished) Similar but less

pro-nounced changes were observed in the more mobile vole Clethrionomys

glar-eolus The occurrence of alternative microhabitats evidently prolongs the

decline phase

These observations imply that folivorous species in homogeneous habitats

or landscapes may show even distributions at large density variations and that the dynamics may be still more violent the more homogeneous an area

or habitat Heterogeneity and clumped distributions may be general indica-tors of population stability, however a prudent proposition to be examined in other systems

Effects of Habitat Juxtaposition

Population Effects

Many species are either dependent on two or more habitats (landscape com-plementation, Dunning et al 1992) or favored by more than one habitat (land-scape supplementation, Dunning et al 1992), both conditions being affected

by the level of connectivity through the intervening matrix (Taylor et al 1993) These dependencies are usually supposed to work close in time, daily, seasonally, or at least annually Spatial scales may vary and examples will be provided later However, long-term effects would also be possible, with far-reaching implications for conservation The proximity of such habitats might affect the type of dynamics within one or several habitat types Interspecific interaction may also influence dynamics within adjacent habitats I will here exemplify these additional effects

A bush cricket, Metrioptera bicolor, studied by Kindvall (1995, 1996) in

south-ern Sweden occurs in subdivided populations in grasslands, surrounded by dry pine forest During most years this species performs equally well on var-ious types of grasslands, but during one particular year (1992), with a long-lasting drought, the crickets survived less well on short-grass grassland and even moved into the forest during the main drought and survived in the shade there Local extinctions would have occurred if the main habitat had not been surrounded by drought-tolerant vegetation Similar observations of spatially varying weather effects have been made on caterpillars of an

Amer-ican butterfly Euphydryas editha associated with serpentine grasslands

(Dobkin et al 1987)

For the cricket M bicolor, habitat heterogeneity, measured as size variability

of various microhabitats in a sampling plot, was negatively correlated with population variability (Kindvall 1996) Populations in some of the most homogeneous plots went extinct during a study period This is one further example that populations in large and fairly homogeneous habitats often are unstable, either with regard to fluctuation patterns or to local persistence Adjacent habitats may show multiannual variations in quality for a certain species A fairly extreme case is the mast seeding of certain deciduous forest groves that are still retained from earlier extensive hardwood forests within the present conifer forest landscape in south central Sweden The extensive

conifer forests support only very sparse populations of wood mice (Apodemus

flavicollis and A sylvaticus (Hansson 1997) At mast seeding in the deciduous

groves there are population outbreaks of granivorous rodents like these

Apo-demus species, but the magnitude of the increase depends on the densities in

the surrounding conifer forests (Hansson 1997), outbreaks in extensive decid-uous forests being much more rapid and reaching higher densities (Pucek et

al 1993) Similar conditions have been observed at artificial feeding with seeds in small patches (e.g., Hansson 1971) The population outbreak in the groves by various small mammal species caused at least one rodent species,

C glareolus, to affect surrounding conifer forest densities by dispersal after

peak food availability

Community Interactions

Under the mast-related population outbreaks, pairs of congeneric species

segregate in space, A flavicollis dominating in the deciduous groves and A

sylvaticus increasing for a shorter period in adjoining conifer forests or other

adjacent environments (Hoffmeyer and Hansson 1974; Hansson 1997)

Simi-lar observations were made for the insectivorous shrews Sorex areaneus (deciduous groves) and S minutus (conifer forest), probably relying on seed

insects (Hansson 1997) In periods without mast production, little evidence of competition is detected (e.g., Hansson 1978)

If a prey or host species to a predator or pathogen, respectively, is affected

by landscape composition then the predator or disease will also depend on this landscape If a predator, pollinator, etc depends on two or more species

of prey, plants, etc in different habitats then landscape composition will again influence predator, etc distributions and probably also numbers Finally, if effects of predation are strong, then alternative prey in another hab-itat may be affected at bottlenecks in the staple food The later conditions are often discussed under the heading of “apparent competition” (Holt 1984)

A case of possible apparent competition (i.e., changing prey) in various habitats is modeled by Hanski and Henttonen (1996) and may explain certain features of Scandinavian vole cycles For instance, this model may be related

to the synchronous dynamics of one rodent species (Microtus oeconomus)

liv-ing on small grassland patches in northern Finland and heavily preyed upon

by stoats and another rodent species (C rutilus) living in adjoining extensive

forests with initially little predation (Heikkilä et al 1994) The grey squirrel

(Sciurus carolinensis) has invaded Great Britain and “driven away” the red squirrel (S vulgaris) in spite of different habitats for these two species

How-ever, a common disease, the grey squirrel being resistant, may possibly have been the agent (Reynolds 1985) Similar disease conditions appear to mediate

“competition” effects of the introduced American crayfish Pacifastacus

lenius-culus on the indigeneous species Astacus fluviatilis in Scandinavia (Gydemo

1996)

Behavior at Edges

Most habitat patches are generally small, particularly with regard to wide-ranging vertebrates, and edge effects are indeed common Different species respond in particular ways to a certain edge and there has been a separation between soft and hard edges (Stamps et al 1987) Many new processes occur

at edges and others are intensified there (Hansson 1994), and populations of various species may respond differently to these changes There has been an emphasis on negative effects (climate, predators, etc.) on populations of for-est interior species (e.g., Temple and Cary 1988), while the use of edges and surrounding habitats by ecotonal species has been considered less

Different movements and dispersal rates of various population categories cause different population compositions at edges and in interior habitat (Gli-wicz 1989; Hansson 1997) At a decline in food resources, many granivorous rodents remained at the edges, exploiting food both in the deciduous grove and surrounding conifer forest In this case, the edges provided the best out

of two worlds (or habitats) for generalized species However, most special-ized species disappear from edges, but remain in larger tracts of interior for-ests (Hansson 1983, 1994 for various bird and mammal species) Similar observations have been made for North American tropical migrant birds (Whitcomb et al 1981) Edges appear generally to be terminators for special-ist species, but refugia for generalspecial-ist species

Animal reactions to edges are, however, not constant In the study on the

cricket Metrioptera bicolor, Kindvall (1995) found the imagoes to cross

grass-land–forest borders extensively in a prolonged drought, a behavior not seen under normal weather conditions These movements were evidently due to physiological or psychological changes Similarly, wood mice invaded mast-seeding groves due to the rich food supply

Effects at Various Spatial Scales

Landscape Complementation or Supplementation at Short Distance

A species may need different resources during daily life or within short time periods These requirements can be fulfilled within a habitat or by move-ments between adjoining or nearby habitats It may sometimes be difficult to distinguish between habitat and landscape effects, as the following examples demonstrate

Passerine birds in the hemiboreal zone show higher mean population num-bers in mixed conifer–deciduous forests than in pure deciduous or conifer forests (Nilsson 1979) Evidently they require food resources, particularly insects and seeds, from the deciduous trees and shelter against predation in the dense but dark and insect-poor conifers Conifers and deciduous trees can grow intermingled (one habitat) or close by (a landscape?)

The Siberian tit (Parus cinctus) of northern Finland prefers habitats with

dead hollow trees, large coniferous trees and insect-rich birches Nesting suc-cess was considerably higher when there was acsuc-cess to these features, even within moderately managed forests (Virkkala 1990) During the fledgling period the tits move to habitats with more birches, and insects, than in the nesting period (Virkkala 1991) As a result of removal of old birches in for-estry, the Siberian tit numbers are now declining regionally

Scandinavian old-growth forests provide habitats for a high biomass of foliose and pendulous lichens These lichens are used as shelter by various insects that are preyed upon by spiders, which in turn are preyed upon by birds (Pettersson et al 1995; Gunnarsson 1996) With each higher trophic level, a wider spatial scale is affected by old-growth remnants

Effects at the True Landscape Scale

Complementation and supplementation are common within the 5- × 5-km scale that is proposed as the genuine landscape scale by Forman and Godron (1986) Most examples come from larger vertebrates, but few other organisms have been examined very carefully in this respect

Scandinavian roe deer forage on fields or clear-cuts, but seek shelter in for-ests, particularly at edges (Hansson 1994) Conditions appear very similar for North American white-tailed deer (Alverson et al 1988) Corvids such as European crows, rooks, and jackdaws feed on fields in winter, but roost in forest groves or city parks (Jonsson 1992)

Such dependencies may strongly affect population numbers; jackdaws can occur in tens of thousands at night in a small town surrounded by open fields (e.g., Uppsala, Hansson, personal observation) Similarly, deer such as moose